🇳🇱
Why the protocol this server is built on (ActivityPub) is the future:
https://blog.joinmastodon.org/2018/06/why-activitypub-is-the-future/
@freemo
> It’s bigger. It’s any piece of software that implements ActivityPub
That for example is not exactly a true statement. ActivityPub is so lax that fragmentation is unavoidable. Of course Eugen can claim such thing because Mastodon happens to be a dominating product that everyone will try to keep compatible with. But not for any minor player
@sandfox ActivityPub is a W3C standard. While you can of course not follow the standard as long as you do then the servers will talk to each other successfully. It has little to do with Mastodon being popular or not.
@freemo What I meant is there may be two programs with correct implementation of activitypub that won't be able to communicate. The stupid example is 2 image sharing apps where one calls its activity "image" and the other one - "picture". Users should be able add each other as friends but their timelines will stay empty. There are infinite possibilities for more subtle metadata incompatibilities
Also I heard from Friendica devs that Mastodon is not strictly compliant but that may also mean that they don't understand the spec, I wouldn't judge here, I don't know
@sandfox I have no issue seeing things in my timelines with a contrary activity type. For example i can subscribe to peertube from ant activity pub server and see all posts, even if i am not on a video specific server. Usually all activity type is shared. Would be against the spec not to
@sandfox PixelFed doing that is not following the standard (though they can do it of course). The proper way to implement that would have been to make the filter user customization so they can set what is or is not in their feed. But otherwise render everything and if the post has a picture in it give it special rendering or something. Obviously assuming a specific activity type for all content is not the intention of ActivityPub
@freemo I think I need to test Mastodon against some edge cases on my spare time. Like feed it an array of objects, unknown types and so on. That's interesting how foolproof is the standard and its current de facto reference implementation